SpaTeoGL: Spatiotemporal Graph Learning for Interpretable Seizure Onset Zone Analysis from Intracranial EEG
Researchers have developed SpaTeoGL, a novel spatiotemporal graph learning framework designed to improve the accuracy of identifying the seizure onset zone (SOZ) from intracranial EEG data. This method constructs window-level spatial graphs of electrode interactions and links them via a temporal graph based on structural similarity. Experiments on a multicenter dataset demonstrated that SpaTeoGL is competitive with existing methods while offering enhanced non-SOZ identification and clearer insights into seizure propagation. AI
IMPACT This graph learning approach could lead to more precise epilepsy surgery by improving the identification of seizure origins.